Design of a
Quadcopter for
Winning the
Jerry Sanders
Creative Design
Competition
Jordan Bramble,
Harold Merida,
Ibtsam Khan
Aurora Flight Sciences
Contents
1.0 Description of Goals of Competition...................................................................................................... 3
2.0 Competition Rules ................................................................................................................................... 8
3.0 Experiment .............................................................................................................................................. 9
4.0 Strategy Analysis ................................................................................................................................... 17
5.0 Mission Requirements .......................................................................................................................... 25
6.0 Functional Requirements ...................................................................................................................... 26
7.0 Design.................................................................................................................................................... 26
7.1 Design Requirements ........................................................................................................................ 28
7.2 Alternatives ....................................................................................................................................... 28
7.3 Weighted Values ............................................................................................................................... 29
7.4 Parts List ............................................................................................................................................ 31
7.5 Pick Up Mechanism ........................................................................................................................... 35
8.0 Simulation Design ................................................................................................................................. 37
9.0 Project Management ............................................................................................................................ 39
9.1 Schedule of Work .............................................................................................................................. 39
9.2 Budget ............................................................................................................................................... 41
9.3 Project and Risk Mitigation ............................................................................................................... 44
9.4 Work Breakdown Structure .............................................................................................................. 47
10.0 References .......................................................................................................................................... 48
2
1.0 Description of Competition
The Jerry Sanders Creative Design Competition, sponsored by Advanced Micro Devices, and
named after their former CEO, is a yearly competition in robotics held at the University of
Illinois Urbana-Champaign. The 2014 competition will be held March 14th – 15th.
Multidisciplinary teams of engineering students nationwide, participate in a game like
competition for two days with robots they’ve designed and constructed. Each year the
competition challenges change and require teams to develop innovative mechanisms for
successfully scoring points. The competition takes place in a 2000 sq. ft. arena (44.7 x 44.7 ft).
For airborne robots, a sloping net is hung above the course that is always approximately 6 ft
above the arena floor.
The goal of the competition is fostering innovation and creativity within robotics and promoting
the engineering disciplines. Each year thousands of spectators attend. In last year’s competition,
26 teams from six different universities participated in the competition, attempting to pick up
cones and place them on pins in order to score points. All of the participating teams hail from
universities within Illinois or Indiana. Historically only two airborne entries have competed.
During last year’s competition, the Northern Illinois University Robotics club entered a
quadcopter and they were successfully able to pick up cones, however they could not accurately
place them on pins and the downward thrust from the propellers would occasionally blow away
unsecured cones before they could pick them up.
Figure 1: An eagle eye view of the course
3
The colored boxes in the corners correspond to each team’s starting area; the four colored
rectangles represent ramps that lead up to the second level territories. At the base of the second
level are small ‘soccer’ balls that are assigned by color to each team, these balls can be moved to
their respective places, around the corner of the second level, in order to change direction of the
ramps. The ramps are equipped with conveyor belts initially moving downward, but changing
their direction causes the conveyor belt to change direction thus pulling a robot up. The central
black box represents the third level territory, and the surrounding 8 boxes represent the second
level. All of the other light blue dots represent the pins in the first level. The light purple boxes
on each edge are designed to hold cones for each team.
Figure 2: Hinged Door Location
The competition shall consist of picking up and transporting cones of a specific color to pins
corresponding to territories, where each pin is 1 inch in diameter. This is equivalent to
controlling a “territory”, and the length of time a territory is controlled determines how many
points are awarded.
Before going into detail about how points are awarded it is necessary to understand where the
territories are placed. Figure 3 is a CAD diagram used to measure the distances between
territories. The type of territory is color coded.
4
Figure 3: Cad Diagram of all the 3 Different Territories
Green represents first level territories, of which there are 36. Yellow represents the second level,
of which there are 8. Finally, there is 1 third level territory marked in red. It is also necessary to
understand how cones are placed in the course.
Figure 4: Pin Location in the Arena
5
For the team starting in the green corner, their cones are arranged in this way. Ten cones are
unstacked behind a drop wall that can be opened by pulling out two pins that hold the wall in
place. Five cones are in front of their home territory, hanging from strings. Five more cones are
placed behind a hinged door beneath the second level that must be pulled open. They also have a
stack of five cones in their home territory as well as 5 cones stacked at the second level and five
opposite of the raised platform. Since the course layout is symmetrical, each other team has their
cones located in equivalent positions throughout the course.
Figure 5: Cones on string
1.1 Scoring
Points are awarded as follows:
•
Airborne robots have a constant 3x multiplier when scoring cones
•
For each contiguous powered territory on the first level controlled at the end of the
match, a team will be awarded ten (10) points
•
For each contiguous powered territory on the second level controlled at the end of the
match, a team will be awarded thirty (30) points
•
For each contiguous powered territory on the third level controlled at the end of the
match, a team will be awarded forty (40) points
•
Every ten (10) seconds that a team controls a first level territory will result in one (1)
point
6
•
Every ten (10) seconds that a team controls a second level territory will result in three (3)
points
•
Every ten (10) seconds that a team controls a third level territory will result in five (5)
points
•
Completion of an action for the first time will result in ten (10) points being awarded to
the team responsible.
An action consists of opening the doors that hold pins or changing the direction of the ramp. The
direction of the ramp can be changed by moving an assigned colored soccer ball at the base of
the second territory to its holding area around the corner of the second level. Clearly, airborne
robots are given a strategic advantage, and additionally, autonomous robots receive a 5x
multiplier.
7
2.0 Competition Rules
The competition can be summarized with the following core rules:
•
Each match will be seven (7) minutes long
•
Each match will consist of four or fewer robots
•
The team’s color cone is the topmost in the stack of cones for control of a territory
•
Teams cannot attempt to control a territory unless it would be contiguous
The following rules are also created for team composition:

A team may have a maximum of six (6) members, consisting of a single team
captain and up to five (5) additional members.

To be eligible to compete, each team member must be a registered student at a
university during the semester during which the competition takes place (Spring),
the preceding semester (Fall), or both.

Only official team members will be allowed in the pit area during the competition
– all others associated with the team will be subject to the same restrictions as
regular visitors.

Other persons besides the official team members may aide in the construction of a
robot, but only the official team members will be recognized at the competition
site.

Optionally a team may have a faculty sponsor, corporate sponsor or a parent as an
advisor in addition to the six official team members. A sponsor is meant to be a
sponsor only. If there is suspicion that a team is abusing the sponsor privilege at
the competition proper, the extra person will be asked to leave. The sponsor is not
an extra team member. The sponsor is allowed in the pit area during the
competition but may not assist in the repair, modification or construction of the
robot.
8
Airborne entries have special specifications. They can weigh no more than 15 lbs. and fit within
a 3 x 3 x 3 cubic feet box. In general, quadcopters are much smaller than this, and weigh much
less. Though building a larger quadcopter should increase stability, since maximizing horizontal
geometry leads to greater stability. Any rotors must be made out of plastic and should not be able
to cut through the safety netting which will be suspended over the course. The same battery rules
apply to airborne entries. Since airborne entries could pose a greater safety risk, teams planning
on submitting an airborne robot must contact the JSDC Rules Chair and Director by December
1st with a description of the expected weight, size, propulsion type, and any other specifications
of their robot which will be helpful in determining what kind of risk such an entry might pose to
the safety of spectators, competitors, JSDC officials and the game course. After the proposal has
been received, the Rules Chair and Director will either approve or deny the entry by January 1st.
In order for the proposal to be approved, it is necessary to consider some of the safety
specifications:

No
explosives,acids,bases,flammables,chemicals,liquids,animals,projectiles,combustibles,
lasers, buzz saws, drills, EMP’s, or unstable power supplies. In summary, nothing that
will endanger the contestants, spectators, officials or the course will be allowed. The
Jerry Sanders Design Competition Committee and Safety Officials have the final
judgment on whether a device constitutes such a danger. Any inquires on the degree of
danger a certain design presents should be directed to a member of the Jerry Sanders
Design Committee.

For public safety and to prevent damage to the course, each robot, whether piloted or
autonomous, must have an easily accessible shut off switch on the robot or be able to stop
on remote
command.
It is also important to consider the types of cones that will be picked up. The cones are Adams
brand cones, approximately 2 inches tall with a 2 inch diameter hole.
Cones will be hanging from strings that span the corners of the course. The strings will be four
(4) feet from the first level of the course and cones will hang one (1) foot down from the strings.
9
(a) The first time a team successfully removes a hanging cone from a string will result in ten
(10) points being awarded to that team.
Robots may not come in contact with other robots, or manipulate or seek to use cones other than
their own corresponding color. Each time these actions are commited, personal fouls will be
assessed. 5 personal fouls results in disqualification, and each personal foul carries -10 points.
Additionally, pilots may not attempt to jam frequencies or interfere with other pilots during
competition. These fouls can result in immediate disqualification.
So, the advantages to using an airborne robot are not only the 3x point multiplier, but also the
speed advantages and the ability to be able to avoid having to use directional ramps in order to
enter higher level territories. In order to verify these advantages, the performance characteristics
of the quadcopter will need to be experimentally verified.
10
3.0 Experiment
In order to later conduct a simulation and also analyze and compare potential strategies for
winning the competition, it is necessary to conduct experimental trials to determine the
performance attributes of a potential quadcopter that would be used in competition.
Three different experiments were conducted. The first measured horizontal velocity over three
separate distances, both taken from measurements on the above CAD diagram, that roughly
corresponds to territory locations. The quadcopter was already hovering when the tests were
conducted. The pilot would fly the quadcopter back and forth between the given distances of
20.5 ft, 30.7 ft, 47.7 ft. each time an observer, observed the quadcopter travel the given distance
the time would be recorded. These times, over 19 trials for each distance, were modeled as
distributions using Arena. The data for these horizontal velocity tests are as follows
11
Table 1: Data for Horizontal Distance
Figure 6: Horizontal Distance Distribution 20.5 ft
Figure 7: Horizontal Distance Distribution 30.7 ft.
12
Figure 8: Horizontal Distance Distribution 47.7 ft
The second experiment measured vertical velocity ascending velocity over 5 different distances,
2.5 feet, 4.5 feet, and 3.5 feet. Due to time constraints in the facility where tests were being
conducted, only 5 trials were conducted for each distance. However the means for each
distribution were very close together. The experiment was conducted by marking the distances
on a wall and mounting a laser to the quadcopter. The laser was aligned with the wall in order to
measure the vertical altitude of the quadcopter. The data is as follows:
Ascend [time]
Trial 1
Trial 2
Trial 3
Trial 4
Trial 5
From 2 Ft
Velocity-ft/sec
2.7
3.3
3.0
1.4
1.9
0.74
0.61
0.67
1.43
1.05
Ascend [time]
Trial 1
Trial 2
Trial 3
Trial 4
Trial 5
From 4.50 Ft
Velocity-ft/sec
5.2
5.0
2.5
2.6
4.9
0.87
0.9
1.34
1.65
0.92
Ascend [time]
Trial 1
Trial 2
Trial 3
Trial 4
Trial 5
5.5 Velocity-ft/sec
3.3
1.67
5.4
1.02
6.7
0.82
3.8
1.45
4.3
1.28
Table 2: Data for Vertical Velocity
13
Figure 9: Vertical Velocity Distribution for fall from 2.5 ft
Figure 10: Vertical Velocity for 4.5 ft
14
Figure 11: Vertical Velocity Distribution 5.5 ft.
The third experiment was testing landing accuracy, in order to determine how accurately the
quadcopter could hover and approach a cone. A target was created and the quadcopter pilot
attempted to land the quadcopter in the center of the target from distances of 2.5, 4.5, and 5.5
feet. Pictures were taken of each landing, and the distances from the center were recorded. 8
trials were recorded from each distance. The data for these trials is as follows:
Table 3: Data Collection for Landing Accuracy
Figure 11: Landing Acc
15
Figure 12: Landing Accuracy Distribution 4.5 Ft.
Figure 13: Landing Accuracy Distribution 5.5 Ft.
16
4.0 Strategy Analysis
It is necessary to analytically evaluate the feasibility of winning this competition with an
airborne robot. In order to do this it is necessary to consider every possible combination of
winning strategies, by developing a programmatic method for determining possible strategies.
These possible strategies can be compared and evaluated via simulation of competition that will
determine the best strategy for competition.
In order to do this it is necessary to have an understanding of the expected aerodynamic behavior
of the quadcopter. For brushless motors, the torque can be calculated as follows:
τ = K t (I − I0 ) , K = torque proportionality constant, I = input current
Where K is the proportionality constant of the motor and I0 is the current when there is no load
on the motor. The Voltage drawn by each motor can be calculated as follows:
V = 𝐼R m + K v ω
Omega is the angular velocity of the propellers, Rm is the resistance of the motor, and K is the
proportionality constant corresponding to back-EMF. We can calculate the Power consumed by
each motor from P= IV, by solving the torque equation for current. If we assume that the motor
resistance and I0 are negligible, and will not be large enough to affect flight time in a noticeable
way, then we can calculate power as follows:
P = (τK v ω)/K t
Additionally, fromP ⋅ dt = F ⋅ dx , we can solve for P = Fv = Tv , where v is hover velocity.
The hover velocity is equivalent to:
T
vh = √
2ρA
17
In this case, rho is the density of the surrounding air, and A is the area, r2, swept out by the
propellers, the thrust, T, when hovering is equal to the weight of the quadcopter plus the weight
of the payload carried. From this, it is clear that larger propellers coupled with thrust being held
constant, equates to less power being drawn from the power supply, and thus higher levels of
efficiency.
Armed with some of the performance capabilities of a standard quadcopter, as well as the
placement of all cones on the course, a strategy for winning the competition can be developed.
First, given the fact that a team can only control a territory that is contiguous to theirs, the
heuristic is that their contiguous path will not be able to intersect with another teams until the
second level at least. Once a team controls, the third level territory then all second level
territories are contiguous to them, and they can then attempt to play defense by placing cones on
top of any territories in the second level that become controlled by the other team. Thus the
optimal strategy is one that allows a team to reach the third level first, and control it as long as
possible. In order to do this, one must consider the placement of the cones on the course, and
pick up cones such that they will have the shortest mean time between cone placements.
The total points at the end of a match can be calculated as follows:
•
F = number of times 1st level territory is controlled for 10 sec.
•
S = number of times 2nd level territory is controlled for 10 sec
•
T = number of times 3rd level territory is controlled for 10 sec
•
X = number of contiguous 1st level territories controlled at end
•
Y = number of contiguous 2nd level territories controlled at end
•
Z = number of contiguous 3rd level territories controlled at end
•
Also add 30 pts for the first time the direction of the ramp is changed, including the 3x
multiplier
18
Then referring back to how points are scored, and incorporating the 3x multiplier, the
total points at the end of the match is:
•
Pts. = 3F + 9S + 15T + 30x + 90Y + 120Z + 30
The constraints for this function are as follows:
•
F≥5
•
S≥1
•
X + Y + Z ≤ 12
We know this since we must control at least 5 territories before we can move into the second
level and at least one in order to move into the third. We calculated an expected number of cones
we can pick up to be 12. So the maximum number of contiguous territories controlled at the end
is X+Y+Z=12.
19
Figure 14: Decision Making Tree
The above diagram illustrates the decision making process that goes into executing a strategy.
This illustrates how the goal is to first cover the 1st level adequately, in order to move into the
second, and then into the third level.
20
Figure 15: Operational Scenario
Figure 15 represents the operational scenario that was composed to represent the decision tree
shown earlier. The interactions are between the user, the quadcopter and the remote controller.
This scenario was created to design the experiments that will follow. The first part of the
scenario involves the user commanding the quadcopter to turn on and ascend. The next steps are
for the user to proceed to the doors and attempt to pick up a cone. In this particular case, the
21
attempt is unsuccessful so the process of attempting to pick up a cone must be repeated until
successfully picked up the cone. The user then commands the quadcopter to proceed to first level
territory and drop the cone into a pin. This entire process of picking up cone and dropping onto
pin is repeated until all contiguous pins are claimed in the first level. The entire process of
picking up cone and dropping into pin is also repeated for the second level territory until all
contiguous territories leading to the third level territory are claimed.
By following the above decision tree, using the distances to each territory mapped out in the
CAD diagram of the arena. We can use the mean times from our velocity distributions to
determine an expected number of territories that can be controlled in a 7 minute period as well as
the number of points that this would attain in a situation where there was no competition.
Refer back to the equation for determining points scored:
•
Pts. = 3F + 9S + 15T + 30x + 90Y + 120Z + 30
•
Mean horizontal velocity of 3.91 ft. sec ( from velocity test)
•
Mean ascending velocity is 1.14 ft. sec (from velocity test)
•
Time to acquire a territory: t = 2d/3.91+ 2*4.5/1.14 +15
This equation for time accounts for the distance there and back motion between territories as
well as ascending and descending. This equation incorporates an assumption that picking up and
dropping off a cone requires 15 seconds. This is optimistic; it may take as long as 45 seconds.
22
Table 4: Expected time to gain control of territories
D is the distance in feet to territories in the arena. The time to gain control is estimated from the
equation above. These distances represent the territories needed in order to control the third
level. It takes 234.1 seconds to control the third level, out of a total of 420 seconds in the match.
The remaining time can be used to play defense and prevent other players from entering the
second level. The mean time to control a territory is 33.4 seconds, which leaves time for 4 more
territories to be controlled in a 7 minute period. Thus, the expected number of cones is 12. In
summary, because territories must be contiguous to other controlled territories, the strategy is to
always approach the nearest territory. Once the third territory has been controlled then a
defensive mode will be assumed, placing cones on top of other teams controlled territories if and
when they enter the second level. Executing this strategy, without any competition taken into
consideration will yield 1908 points.
A mean time of 33.4 seconds to control a territory may be too optimistic. If it takes 30 seconds to
pick up and drop a cone, then the mean time to control a territory is 48 seconds, which is enough
to control 8 or 9 territories. This means the third level can be controlled, with two cones
remaining to use defensively. In this case, the final point tally, not considering competition is
1,281.
23
Figure 16: Graph of Points versus Time with No Competition
24
5.0 Mission Requirements
1. The Quadcopter shall compete in the Jerry Sanders Creative Design Competition.
2. A strategic support tool shall be developed to determine and verify optimal strategies for
winning the competition.
3. The Quadcopter shall have a mechanism to pick up and carry cones, in order to place
them on territories throughout the course.
25
6.0 Functional Requirements

The Quadcopter shall have the ability to transmit a live video feed to a laptop.

The Quadcopter shall be able to remain in flight for at least 7 minutes at a time. A 3
minute buffer is desirable.

The Quadcopter’s batteries shall be rechargeable between rounds.

The batteries shall provide sufficient power such that, the flight time and consistent thrust
to carry necessary payloads can be achieved.

The Quadcopter shall lift the cone to a sufficient height, at least 2 feet above the ground.

The Quadcopter shall place cones on a 1 inch diameter 8 inch long pin.

The Quadcopter shall have the ability to collect cones semi-autonomously.

The Quadcopter shall be able to fly beneath a 6 feet net at all times, and have the ability
to fly as low as 2 feet for an extended period, totaling no more than 20 seconds.

The Quadcopter shall have a carrying capacity of at least 1kg, for a camera, cone
acquisition mechanism, and cone.

The Quadcopter shall be transportable, able to meet requirements to be checked onto a
plane, or in the trunk of a sedan.

The Quadcopter’s Arduino system shall be compatible with stability (fly-by-wire)
software modules.
26
Functional Architecture
27
7.0 Design
7.1 Design Requirements
•
The Quadcopter shall fit within a 3’x3’x3’ cube.
•
The Quadcopter shall weigh no more than 15lbs.
•
The Quadcopter shall be equipped with a FPV camera.
•
The Quadcopter shall have a mechanism for picking up Adams Saucer cones with a 2’’
diameter central hole and 2’’ height.
•
The Quadcopter’s propellers shall be guarded such that they cannot damage the netting
above the course.
•
The Quadcopter shall be equipped with a mechanism for concentrating the thrust such
that it will not blow cones away when attempting to pick them up.
7.2 Alternatives
Figure 17: Alternate Quadcopter
Three different options for Quadcopter kits are in consideration, which will likely need to be
heavily modified. The first is manufactured in China and sold by the DC Area Drone Users
group. It is essentially identical to the 3D Robotics model, but has a cheaper frame and may not
provide the stability that we are looking for. The second model is the A.R. Parrot Drone. This
drone offers a very long flight time and is built on an open source android Operating System,
however it cannot carry anywhere close to the payload that will be necessary for this
competition. The 3D Robotics ArduCopter, though more expensive, offers a reliable shipping
28
time and stable frame. It also can carry 60% more than the necessary payload, and is highly
customizable.
7.3 Weighted Values
29
Caregory Weight: Score (0-5):
Score:
Carrying Capacity:
DC Drones
A.R. Parrot
3D Robotics
Flight Time With Max Payload:
DC Drones
A.R. Parrot
3D Robotics
Quadcopter Weight:
DC Drones
A.R. Parrot
3D Robotics
Technology:
DC Drones
A.R. Parrot
3D Robotics
Cost:
DC Drones
A.R. Parrot
3D Robotics
Camera:
DC Drones
A.R. Parrot
3D Robotics
1600 g
250 g
1600 g
5
5
1
5
25
5
25
11.5 mins
0 mins
11.5 mins
5
5
0
5
25
0
25
1000 g
420 g
1000 g
4
3
4
3
12
16
12
Arduino
Linux
Arduino
3
4
2
4
12
6
12
$1,700
$300
$1,800
3
3
4
2
9
12
6
520p
720p
520p
3
4
3
4
12
9
12
Figure 18: Trade off Analysis
The cones are 2” Adams saucer cones, testing scenarios will be developed to compare each
technology option quantitatively to determine their precision (accuracy of picking up cones). The
methods will first be tested in place without being attached to a moving quadcopter. They will all
be compared with the same amount of trials. If any of the methods are able to achieve at least an
80% level of accuracy then they will also be tested on the quadcopter in flight.
The weights were chosen based upon the nature of the competition. Flight time and carrying
capacity are highly important, and keeping weight of the robot down is also important to ensure a
lengthy flight time. The abilities of the technology such as autopilot and camera are less
30
important as they are only needed in this competition for lower level tasks, and most of the flight
time will completed manually. Because all options are reasonably priced, cost is only of medium
importance to us. The DC drone user groups, imitation ArduCopter is clearly the best choice.
7.4 Parts List
Table 5: Camera Specs
Also being considered: using a Camera, for manually guiding the cone acquisition mechanism. The Go
Pro is the most popular configuration, but it is heavy. Using a cheaper and lighter 3DR FPV transmission
is preferable.
KV850 Motor
10 " Propeller
Power
Supply
3300 mAh 30/45C
Flight
time
11" Propeller
4000
4500
4000
4500
mAh
mAh
mAh
mAh
30/45
30/45
30/45
30/45
C
C
C
C
3300 mAh 30/45C
11.4mi 11.7mi
9.5min
n
n
14.6mi
10.8min
13min
n
Mixed
Flight
time
15.4mi 16.2mi
12.8min
Hover
Flight
n
n
19.1mi
14.1min
37.7mi 44.2mi
31.1m
n
n
17min
n
36.7mi 41.3mi
30.2min
n
n
31
time
Battery
Weight
558g
678g
762g
558g
678g
762g
Efficienc
y
67.3% 67.6% 67.7%
71.4% 71.7% 71.8%
Payload
Capacity
1356g 1380g 1394g
882g
897g
906g
Table 6 Alternative Camera Kits
KV880 Motor
10 " Propeller
Power
Supply
3300 mAh 30/45C
11" Propeller
4000
4500
4000
4500
mAh
mAh
mAh
mAh
30/45
30/45
30/45
30/45
C
C
C
C
3300 mAh 30/45C
Flight
time
10.3mi
6min
7.2min 8min
7.7min
9.2min
n
Mixed
Flight
time
10.5mi 11.7mi
8.8min
n
n
12.8mi 14.3mi
10.7min
n
n
Hover
Flight
time
41.6mi
34.6mi 38.9mi
30.5min
37min
n
28.5min
n
n
558g
678g
762g
558g
678g
762g
Battery
Weight
Efficienc
y
67.3% 76.7% 67.7%
79.3% 79.7%
80%
1549g 1579g 1596g
1035g 1054g 1065g
Payload
Capacity
Table 7: Brushless Motor Alternatives
32
Two different Motors (KV 880 and 850) are being considered with configurations of 10 inch and
11 inch propellers allow with 3 different power supply simulations. Ideally a simulation and test
plan of all 12 of these configurations will be conducted to see which yields the best flight time
for the desired payload. The issue is that, Quadcopters with higher thrust can carry higher
payloads, but have shorter flight times. Larger power supplies can give a better flight time, but
also add more weight. This creates a catch-22 situation.
Part Name
Cost
Function
Motor Type: 3DR multicopters come standard with 850 Kv
motors and 10" propellers, we upgraded to 880 Kv motors and
Motor 880Kv
$96
11" inch propellers for even more power.
It allows the user to turn any fixed, rotary wing or multi rotor
vehicle (even cars and boats) into a fully autonomous vehicle;
capable of performing programmed GPS missions with
APM 2.6 Autopilot
System
$159.99
waypoints. Available with top or side connectors.
The APM 2.5 Power Module is a simple way of providing your
APM 2.5 with clean power from a LiPo battery as well as
3DR Power Module
current consumption and battery voltage measurements, all
with XT60-Deans
through a 6-pos cable. The on-board switching regulator outputs
Battery Adapter
$99.96
5.3V and a maximum of 2.25A from up to a 4S LiPo battery.
APC Propeller pack. Packet includes one Pusher and one
11" Propellers
$31
Normal plus adapter ring for each. APC 11x47 SFP Style.
This new design incorporates the HMC5883L digital compass,
providing a convenient method of mounting the compass away
from sources of interference that may be present in the confines
3DR ublox GPS with
Magnetometer
$89.99
of the vehicle.
This new ArduCopter frame design incorporates several new
Quad Copter Frame
$99.99
features. The landing gear is moved further apart making the
33
vehicle more stable during takeoff and landing. It provides a
wide unobstructed field of view for mounting cameras under the
vehicle.
Replacement Kit: added a replacement kit to get back in the air
fast. The replacement kit includes two motors, two propellers,
two arms, two legs, two electronic speed controllers, spare bolts,
Replacement Kit
$114.99
and extra spacers.
Telemetry Set: highly recommended adding a 3DR Radio
Telemetry Kit to your autonomous vehicle. Telemetry radios
allow your ground station computer to communicate with your
aircraft wirelessly, providing unparalleled ease of use for
viewing real-time flight data, changing missions on the fly,
tuning, and using an Android tablet as a ground station. Select
915 MHz for the US and 433 MHz for Europe. Arrives preconfigured with RTF copter! The telemetry set includes an air
module, ground module (USB connector), telemetry connector
Telemetry Set
$85.99
cable, two antennas, and a USB extension cable.
Features-Extreme low output resistance-Multiple protection
features: Low-voltage, cut-off protection / over-heat protection3 start modes: Normal, Soft, Super-soft -Throttle range can be
programmed-Smooth, linear and precise throttle response.Separate voltage regulator IC for onboard microprocessorSupported motor speed (Max): 210000 RPM (2 poles), 70000
20 Amp ESCs with
SimonK firmware
$103.96
RPM (6 poles), 35000 RPM (12 poles)
Radio Control: Multicopters require a radio control system to be
used as either primary or backup control. If you don't already
have an RC transmitter and receiver, Spektrum DX8 will
Radio Control
$429.99
connect and configured to meet the requirements.
OSD/FPV: Selected the 3DR OSD/FPV Kit to stream live video
from your APM: Copter with superimposed live telemetry data.
Great for an immersive flying experience with complete in-theair data. This kit includes a Sony HAD 520 line camera,
FPV Camera +
Transmitter
$249.99
MinimOSD on-screen-display module, 5.8 GHz video
34
transmitter and receiver pair, two 900 mAh LiPo batteries, and a
specially designed Y-cable to allow you to use your Telemetry
Radio and OSD using the same APM telemetry port. The 3DR
OSD/FPV Kit comes pre-configured with RTF copter!
Battery Kit: 3DR multicopters are powered by rechargeable
lithium polymer (LiPo) batteries. Selected this option to receive
a battery kit that includes one 3S 4200 mAh 11.1v LiPo battery,
a LiPo balance charger, and a LiPo bag for safely charging and
transporting the battery. We are receiving three batteries with
Batteries
$137.99
our kit.
For testing the various mechanisms for 'cone acquisition'
mentioned in requirements, buying cones identical to the ones
$300
sizes and motor configurations.
R & D parts
Total Cost:
used in competition, 3D printing parts, testing various propeller
$1999.84
Table 8: Parts List Cost
7.5 Pick up Mechanism
A mechanism is needed for picking up cones and transporting them within the arena. Currently
there is no precedent, as previous airborne contest entries did not have much success with
picking up cones, so it is necessary to test numerous options. The first option being considered is
a remote 3-piece claw with high friction anti-slip surface that can be closed on the surface of the
cone and can be guided into place with an FPV camera. It is possible that the quadcopter will not
fly steady enough to manually pick up cones with a high level of precision, so the other option is
to integrate open source autonomous color detection software with the microcontroller to
automatically track the cone and grab it autonomously, once the pilot has manually positioned it
35
within certain vicinity. The third option being considered is a “hook” that can be inserted into the
central hole of the cone and then expanded. The issues with this are that it may not be reasonable
to expect the quadcopter to be able to fly with the precision necessary to center the quadcopter;
additionally it may not be possible to unstack cones that are stacked. The budget for testing and
developing different configurations has been included in the R&D budget.
36
Figure 19: Figure of Pick up Mechanism Possibilities
8.0 Simulation Design
As outlined in the strategy analysis section, since a quadcopter will likely need to ignore the
cones that are behind the doors or hanging from strings, so the quadcopter must focus on the
three stracks of cones that are available to them. One thing that is true for all competing teams is
37
that they must only control territories that are contiguous with ones they already control. So
heuristically, it can be determined that the only instances where this can occur, is when the
contiguous shape created by one team approaches a territory that’s already controlled, or
contiguous with a line of territories controlled by an opposing team. Since the most points come
from controlling territories in the second and third level, the only place where this is likely to
happen is inside of the second level. The team that will have the ability to play defensively is a
team that is already controlling the single third level territory, since at that point all other second
level territories are contiguous, and they can prevent other teams from controlling anything in the
second level. Hence, the “name of the game” is to move into and control the third level as soon
as people, hopefully first, and use the remaining time to place cones onto opponents controlled
territories in the second level in order to prevent them from taking over the third level.
Since the rules state that a team can only control a territory contiguous to a territory they already
control, this means the first territory a team controls must be contiguous to their home territory in
the first level. In this case, it was previously illustrated in the that at least 5 territories must be
controlled in the first level, and one in the second before the third can be controlled.
Choosing where to pick up cones on the course can greatly affect the time to control territories.
For a quadcopter, the ideal strategy would be to immediately proceed to the nearest stack of
cones after controlling a territory. For a ground robot that can open the doors, in instances where
the nearest cones are the ones behind doors, picking those up may be beneficial.
Now that experiments have been conducted, in order to model horizontal velocity, landing
accuracy, and vertical take off velocity, these distributions will be used in the simulation trials,
alongside the distances of cones and territories to be controlled in order to determine expected
number of territories that can be controlled, and output the number of points. The landing
accuracy distributions will factor in to determining error potential for picking up cones as well as
estimating the time to pick up a cone.
The strategic support tool will be designed using an object oriented language, with a robot class
that takes parameters, for max velocity, agility, user error, and precision of picking up cones.
38
During the simulation, user error and precision can be combined with a random number to
determine whether or not a cone is picked up or not. These stats can be used to determine a
maximum amount of cones a robot can pick up.
Since a quadcopter will only be picking up cones that are in one of the three stacks, the time to
pick up and drop off a cone will be the same. However for competing robots, it will take
different lengths of time to pick up cones on strings and behind doors. Accurate estimations of
this time will need to be factored into the simulation.
Simulation trials will be ran against 4 competitors, and all of their possible choices for cone
placement and choices of cones to pick up, in order to see the outcomes. In the simulation,
whichever team is currently in control of the third level territory will assume a defensive role in
order to prevent the other team from controlling the third level.
Ideally this simulation will be conducted using MATLAB, and come complete with a GUI for
user interaction and visualization of territories controlled by each team, as well as which cones
they selected. Though if necessary, for web optimization, it could also be written in a more web
friendly scripting language such as Python. The output will show the CAD diagram with
highlighted controlled contiguous territories, as well as selection of cones.
9.0 Project Management
9.1 Schedule of Work
The project schedule below shows the tasks and their expected durations.
39
Figure 20
Figure x shows the schedule of work and Gantt chart for the entire project. The Critical Path is
marked in red. A critical path can be defined as a “sequence of stages determining the minimum
time needed for an operation.” The tasks that are part of the critical path include: the context
40
analysis, simulation design, mission requirements, modifications based on project briefing 1,
physical construction of the quadcopter, testing and flying of the quadcopter, and all preparations
of the final papers and presentations. Tasks that lie on the critical path need to be closely
monitored to avoid any delays in the project.
9.2 Budget
There were a couple of budgets that were calculated for this project. These estimations are based
on a salary of $55 per hour. An overhead price has to be adjusted since George Mason delegates
53 cents out of every dollar an employee charges their customer bringing the total to $117 per
hour. The three budgets that were composed were composed by having a 10% overall budget
expense and a 90%. The cost of the quadcopter itself was included as well as travel expenses to
the Jerry Sanders Design Competition in Chicago. The blue line in the figure is the predicted
total cost of the project, parts and labor. The hours worked was determined by predicting the
number of hours each individual employee planned on working on the project. The red indicates
the actual cost incurred up at this and will be updated every project briefing to show progress. At
this point in time, no money has been spent on the parts; the majority of the cost has been labor.
The total cost for all three budgets are as follows: $125,000 for planned cost, $137,000 for the
worst case, and $113,000 for the best case. The following figures are breakdowns of the salary
(table 9), conference costs (table 10), and all the calculations for the budgets (table 11).
Team Member Hours
Jordan
Harold
Ibby
Overhead
Hourly Rate Total Hourly
2.128
$55.00
$117.02
2.128
$55.00
$117.02
2.128
$55.00
$117.02
Table 9: Hourly Wages
Travel Conference Expenses Predicted
Transportation of Quadcopter
Flight Tickets
Hotel
Food
$150.00 UPS Overnight
$1,527.00 509/ person via United
$300.00 $100 / night
$150.00
Table 10: Travel Expenses
41
Week
Jordan
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
Total Labor Cost
Total Parts Cost
Travel To Conference
TOTAL ESTIMATED COST
Harold
22
18
19
22
22
17
15
16
12
10
14
16
20
22
14
12
11
9
8
4
8
7
10
9
4
3
4
4
5
6
3
2
5
4
3
5
6
2
Ibby
13
15
12
12
18
9
8
10
12
11
15
12
12
11
8
12
13
7
7
5
9
8
11
8
3
5
5
5
3
4
5
3
2
6
4
4
5
2
Total
11
10
9
10
16
11
10
10
12
10
9
14
12
18
15
14
10
6
9
7
7
9
12
6
5
6
5
3
7
3
4
5
3
4
2
2
7
4
46
43
40
44
56
37
33
36
36
31
38
42
44
51
37
38
34
22
24
16
24
24
33
23
12
14
14
12
15
13
12
10
10
14
9
11
18
8
Planned Cost
PlannedCost WORST Case
Worst Cumulative
$5,382.98
$5,382.98
$5,921.28
$5,031.91 $10,414.89
$5,535.11
$4,680.85 $15,095.74
$5,148.94
$5,148.94 $20,244.68
$5,663.83
$6,553.19 $26,797.87
$7,208.51
$4,329.79 $31,127.66
$4,762.77
$3,861.70 $34,989.36
$4,247.87
$4,212.77 $39,202.13
$4,634.04
$4,212.77 $48,041.89
$4,634.04
$3,627.66 $51,669.55
$3,990.43
$4,446.81 $56,116.36
$4,891.49
$4,914.89 $61,031.26
$5,406.38
$5,148.94 $66,180.19
$5,663.83
$5,968.09 $72,148.28
$6,564.89
$4,329.79 $76,478.06
$4,762.77
$4,446.81 $80,924.87
$4,891.49
$3,978.72 $84,903.60
$4,376.60
$2,574.47 $87,478.06
$2,831.91
$2,808.51 $90,286.57
$3,089.36
$1,872.34 $92,158.91
$2,059.57
$2,808.51 $94,967.43
$3,089.36
$2,808.51 $97,775.94
$3,089.36
$3,861.70 $101,637.64
$4,247.87
$2,691.49 $104,329.13
$2,960.64
$1,404.26 $105,733.38
$1,544.68
$1,638.30 $107,371.68
$1,802.13
$1,638.30 $109,009.98
$1,802.13
$1,404.26 $110,414.23
$1,544.68
$1,755.32 $112,169.55
$1,930.85
$1,521.28 $113,690.83
$1,673.40
$1,404.26 $115,095.09
$1,544.68
$1,170.21 $116,265.30
$1,287.23
$1,170.21 $117,435.51
$1,287.23
$1,638.30 $119,073.81
$1,802.13
$1,053.19 $120,127.00
$1,158.51
$1,287.23 $121,414.23
$1,415.96
$2,106.38 $123,520.62
$2,317.02
$936.17 $124,456.79
$1,029.79
$119,829.79
$131,812.77
$2,500.00
$2,500.00
$2,127.00
$2,127.00
$124,456.79
$136,439.77
$5,921.28
$11,456.38
$16,605.32
$22,269.15
$29,477.66
$34,240.43
$38,488.30
$43,122.34
$47,756.38
$51,746.81
$56,638.30
$62,044.68
$67,708.51
$74,273.40
$79,036.17
$83,927.66
$88,304.26
$91,136.17
$94,225.53
$96,285.11
$99,374.47
$102,463.83
$106,711.70
$109,672.34
$111,217.02
$113,019.15
$114,821.28
$116,365.96
$118,296.81
$119,970.21
$121,514.89
$122,802.13
$124,089.36
$125,891.49
$127,050.00
$128,465.96
$130,782.98
$131,812.77
Best Total
$4,844.68
$4,528.72
$4,212.77
$4,634.04
$5,897.87
$3,896.81
$3,475.53
$3,791.49
$3,791.49
$3,264.89
$4,002.13
$4,423.40
$4,634.04
$5,371.28
$3,896.81
$4,002.13
$3,580.85
$2,317.02
$2,527.66
$1,685.11
$2,527.66
$2,527.66
$3,475.53
$2,422.34
$1,263.83
$1,474.47
$1,474.47
$1,263.83
$1,579.79
$1,369.15
$1,263.83
$1,053.19
$1,053.19
$1,474.47
$947.87
$1,158.51
$1,895.74
$842.55
$107,846.81
$2,500.00
$2,127.00
$112,473.81
Table 11: All Budgets
42
Figure 10: All Predicted Budgets for 38 weeks
Figure 21
43
The Cost Performance Index (CPI) and Schedule Performance Index (SPI) for the project are
calculated for each working week. A ratio closer to 1 is ideal and it shows whether or not the
team is ahead in schedule or behind. It also states whether or not the team is under budget or
over budget for the current week. An SPI greater than 1 is good meaning the project is ahead of
schedule. A CPI less than 1 means the cost of completing the work is higher than planned, a CPI
of 1 means the cost of completing the work is right on plan and greater than 1 means the cost of
completing work is less than plan. A CPI greater than 1 is not necessarily a good thing as it
shows that the initial plan was too conservative.
9.3 Project and Risk Mitigation
Risk
Damage during testing
Mitigation

3D print parts for low level test
configuration
Damage during competition
Damage during transportation of quadcopter

Adequate practice

Purchase spare parts

Budget for spare parts

Adequate practice time

Transport quadcopter using a reliable
carrier

Drive to competition and ensure delivery
of quadcopter
Incorrect simulation assumptions are made

Ensure assumptions used in the model are
valid by analyzing prior competition results

Involve as many of the stakeholders as
possible.
Operator error

Ensure enough practice occurs

Practice with a small scale quadcopter

Select the best operator
44
Computer Error (if autonomous route is

pursued)
Ensure to the best of ability that no error is
made

Create operational scenarios to ensure
that all scenarios are considered.
Hardware Failure at any time
Going Over Budget
Time Constraint
Loss of essential Quadcopter Parts

Revert back to manual control

Extensive testing

Purchase extra parts

Ensure adequate practice time

Perform adequate analysis

Follow CPI and SPI charts

Start earlier

Work efficiently

Actively monitor project schedule

Keep an active inventory of all active parts

Purchase of additional parts
Table 12: Risk Analysis Table
Table 8 is a table of risk analysis about potential risks that could arise at any point during the
project and how to mitigate them.
Damage during test: Having access to a 3D printer makes it easier to print out parts cheaply.
Bringing the cost down and allowing more testing to be done. A lot of the parts on the
quadcopter are interchangeable which make this a very feasible way to mitigate this risk.
Adequate flight time and practice will ensure that the probability of a crash and possible damage
during testing is reduced.
Damage during competition: Purchasing additional parts will help in the case of damage during
competition. It isn’t possible to use the parts that were 3D printed because it would influence the
flight and other variables that shouldn’t be changed.
45
Damage during transportation of quadcopter: Transportation of the quadcopter is something
that was thought over a lot. There is a high probability of damage during transportation as
nobody is sure whether or not carriers treat fragile items as such. Some ways to mitigate this risk
is to either transport quadcopter using a reliable carrier which has some risk in itself, or a more
feasible possibility would be to drive to the competition and that would ensure that our
quadcopter makes it there in one piece.
Incorrect assumptions are made: It is natural that there may be some incorrect assumptions
made when thinking of an effective strategy. To mitigate, we ensured the assumptions that were
used in the model were as correct as we could make them, another way was to involve as many
stakeholders as possible to ensure that the assumptions that were made were correct.
Operator Error: There is always a risk of operator error, whether it is in practice or during
competition. Initial testing with a small scale quadcopter will help determine the best operator
for the competition. Adequate practice time will ensure automaticity in flying the quadcopter
minimize the risk of an accident during competition.
Computer Error: It is very difficult to determine whether or not a computer error will happen if
the autonomous route is chosen, but to reduce the risk the following things can be done: ensure
to the best of the teams ability that no error was made when programming the microcontroller,
create operation scenarios about a few possible scenarios where things can go wrong. Another
feasible way to mitigate this risk is to revert back to manual control if any malfunction occurs.
Extensive testing of onboard computer will minimize the risk of failure.
Hardware failure at any time: Hardware failure can happen at any time during the project. To
reduce this risk, it is necessary to purchase extra parts. With more test flights, hardware
irregularities will become apparent.
Going Over Budget and Time Constraint: All projects face a risk of going over budget and
being behind in schedule. To mitigate these, it is important to perform adequate analysis to
ensure all the parts that are being considered for purchase are essential. Another way is to follow
the Cost Performance Index and Schedule Performance Index charts to ensure cost and schedule
are kept at bay. To reduce time, it is important to start work earlier and efficiently as well as
actively monitoring the project schedule at all times.
46
9.4 Work Breakdown Structure
Figure 22 Work Breakdown Structure
47
10.0 References
1. www.3drobotics.com
2. www.diydrones.com
3. JSDC Rules List: http://jsdc.ec.illinois.edu/news/2013/01/jsdc-2013-competition-rules
4. Quadcopter, Dynamics, Simulation, Control - Andrew Gibiansky- for equations for
simulation design
48